I’m not sure that it’s a matter of them not rushing to release, but that they simply can’t get it out there any sooner given the allegedly huge issues they are facing with things like die size, thermals and lets face it, potential IP issues which I guarantee Apple has all of their legal/IP team watching with a magnifying glass.
I also think Qualcomm doesn’t have a lot of choice to do anything else as every analysis I’ve read says that the current Snapdragon design is rapidly running out of headroom.
That being said, I genuinely hope they succeed as the market and users will benefit. It’s just that I think that many are underestimating the size and number of hurdles they face in the hope of an alternative to Apple.
BTW, we hear that the moment that any Nuvia based hardware gets outside of Qualcomm, several ARM vendors (including a graphics card maker and not Apple surprisingly) are going to be ready with IP related lawsuits. TLDR, the legacy of what the Nuvia engineers did for Apple, may also be their biggest anchor.
You guys…I’m trying to build a little hype here, and everyone is piling on QC.
Well I for one will not be daunted. It could be worse; they could have rushed it out the gate half-baked just to please shareholders for the Q4 sales season. Sure there would be problems, but that’s for Q1-Q2 quarterlies to clean up, and Amon could say he hit all targets for 2023, etc.
Lawsuits never daunted QC either. No, Qualcomm’s going to give it their all for long-term success. Amon’s probably got his entire chair riding on this.
(Hifihedgehog - Waiting for Surface Pro 10!)
SemiAccurate previously told you about Qualcomm/Nuvia’s Oryon SoC that taped out late last year. What we didn’t tell you about is the projected performance and how it is doing now.
Lets start out by saying that as of mid-year the prospects were positive, the SoC had severe issues but those were expected to be cured with a stepping that was in the oven as of Q2. We will ignore Qualcomm’s veiled promises of Oryon being a laptop and phone part which subtly morphed into just being a laptop part for now, it wasn’t by chance.
As we know by now the SoC is the Nuvia core force welded into the Qualcomm uncore, and that is where many of the problems lie. It is a server core and was always meant to be a server core, not a consumer one. Power management is one of the key differences there, and yes we know about the power management die ala Apple’s Monaco but still aren’t clean on the patent situation some have mentioned to us. In short there are a lot of things you can add to a core late in the game, power management is not one of them.
Back to the story at hand we have the now laptop only Qualcomm Oryon part. The core is fine, it hit performance and power targets more or less, and that is good. If you haven’t heard by now, performance is said to be a little better than an Apple M2 and should debut, err, after Apple’s M3. On a brighter note, power is, umm, well it won’t see phone use this generation but somehow this will be spun into a win.
Back to the core/uncore from two different IP ancestries. This is the cause of the major problems that have been dogging Qualcomm for a while now. As of Q2 the device was, umm, not good but the next stepping is intended to cure everything. It is back so how are things now?
Why is Nuvia focusing on this 1W-4.5W per core power envelope?#
I’ll let Nuvia answer this:
All current and future flagship server SoCs are power constrained, very much like mobile SoCs. As core count increases, what is not increasing is the TDP. …The TDP range is 250W — 300W, and the power outside of the CPU can range between 10W — 120W depending upon the workload. Taking into consideration these factors, the amount of power that each CPU can be allocated ranges between 1W — 4.5W when heavily utilized, as is the case in a datacenter environment.
So the major challenge is scaling down the architecture to 8 P-cores and lower. It sounds like they were planning on solving this by integrating four Cortex-A510 E-cores from their Kryo architecture, but ran into issues with the power management system (PMICs).
So this would probably explain why QC is so adamant about manufacturers using Qualcomm PMICs, as earlier noted by @dstrauss.
Yes. to their credit, Qualcomm is trying a genuinely new approach to the whole complex puzzle that is TDP versus cores.
Apple very much has relied on a few performance cores to do most of the work (to good effect in benchmarks) while having possibly the downside of not actively encouraging developers to better balance their apps to use all the aspects of the processor.
And @Marty that’s one of the areas I’m optimistic about Hamoa specifically as it relates to WOA as Windows scheduler (despite my occasional jabs at aspects of it) is far more flexible and accessible to developers, though alas, the number of developers to date that actually have done so usually could be counted on Homer Simpsons right hand.
And, one of the biggest things Qualcomm also needs to overcome is an IOS/A series first approach among app developers that even tried to build WOA native apps.